A. Field of the Invention
The present invention relates to constant voltage reference circuits. More particularly, the present invention relates to a bandgap voltage reference circuit wherein (i) the output voltage can be low and set relative to the silicon bandgap voltage, (ii) the output voltage can have a zero TC, and (iii) the operating supply voltage Vcc can be limited.
B. Description of the Related Art
So-called bandgap reference circuit produces an output voltage that is approximately equal to the silicon bandgap voltage of 1.206 V (hereinafter termed simply the xe2x80x9cbandgap voltagexe2x80x9d) with a zero temperature coefficient (xe2x80x9cTCxe2x80x9d).
1. FIG. 1xe2x80x94Prior Art
FIG. 1 shows a prior art bandgap reference circuit, sometimes called the Brokaw bandgap circuit. This circuit is built with current sources I1-I2, npn bipolar junction transistors Q1-Q2, resistors R1-R2, and operational amplifier (xe2x80x9copampxe2x80x9d) A1. Opamp A1 has a negative input terminal (node n1), a positive input terminal (node n2), and an output terminal (node n3).
Current sources I1-I2 are implemented so that each current source produces a substantially equal current I. This can be done, for example, by utilizing p-channel MOS transistors. In such an implementation, the source of each PMOS transistor is connected to Vcc, and the gates of the PMOS transistors are connected together in a current mirror configuration to node n1.
Transistor Q2is N times larger in size than transistor Q1. Initially, with Q2 larger than Q1 and equal current from I1-I2, the voltage across Q1 will be N times larger than the voltage across Q2. Thus, node n1 will be driven higher than node n2. This will cause the voltage at node n3 to increase. The bases of transistors Q1 and Q2 are connected to node n3, so increasing the voltage at node n3 causes current I from current sources I1-I2 to increase. Current I will increase until the voltage across resistor R1 balances the voltage difference between transistors Q1 and Q2.
The equilibrium value for the current I is given by                     I        =                              Δ            ⁢                          xe2x80x83                        ⁢                          V              BE                                            R            1                                              (        1        )            
The difference in the base-emitter voltage of the two transistors Q1 and Q2 is expressed as                               Δ          ⁢                      xe2x80x83                    ⁢                      V            BE                          =                              kT            q                    ·                      ln            ⁡                          (              N              )                                                          (        2        )            
Because xcex94VBE is a function of thermal voltage kT/q, it is said to be proportional to absolute temperature (PTAT).
The output voltage Vout1 in FIG. 1 is expressed as                               V          out1                =                              V                          BE              1                                +                                                                      2                  ·                                      R                    2                                                                    R                  1                                            ·              Δ                        ⁢                          xe2x80x83                        ⁢                          V              BE                                                          (        3        )            
Three observations can be made about Vout1. First, for a certain ratio of the resistors R1 and R2, Vout1 becomes equal to the silicon bandgap voltage. Second, Vout1, does not depend on the absolute value of the resistors used, which is hard to control. Third, Vout1 is temperature independentxe2x80x94that is, it has a zero TC.
B. FIG. 2xe2x80x94Prior Art
Most modern CMOS processes have only substrate pnp bipolar junction transistors available. In this case the collector of the pnp transistor is forced to be the VSS/ground node. The configuration for a bandgap reference circuit using this type of bipolar junction transistor is shown in FIG. 2.
The circuit of FIG. 2 is built with current sources I3-I5, pnp bipolar junction transistors Q3-Q5, resistors R3-R4, and opamp A2 Opamp A2 has a negative input terminal (node n4), a positive input terminal (node n5), and an output terminal (node n6).
Current sources I3-I5 are implemented so that each current source produces a substantially equal current I. As described above, this can be done by utilizing PMOS transistors.
Transistor Q4 is N times larger in size than transistors Q3 and Q5. Initially, with Q4 larger than Q3 and Q5 and equal current from I3-I5, the voltage across Q3 and Q5 will be N times larger than the voltage across Q4. Thus, node n4 will be driven higher than node n5. This will cause node n6 to increase, causing the current I from current sources I3-I5 to increase. Current I will increase until the voltage across resistor R3 balances the voltage difference between transistor Q4 and transistors Q3 and Q5.
In this case, the output voltage Vout2 in FIG. 2 is expressed as                               V          out2                =                              V                          BE              5                                +                                                                      R                  4                                                  R                  3                                            ·              Δ                        ⁢                          xe2x80x83                        ⁢                          V              BE                                                          (        4        )            
As with Vout1 in FIG. 1, Vout2 can be set equal to the silicon bandgap voltage, Vout2 is temperature independent, and Vout2 does not depend on the absolute value of the resistors used.
The prior art circuits of FIGS. 1 and 2 cannot work with supply voltages below about 1.5 V, since the bandgap voltage with a zero TC is about 1.2 V for silicon. Many applications, however, require the voltage reference circuit to operate with a voltage supply below 1.5 V. The present invention presents such a circuit.
In accordance with the present invention, a bandgap voltage reference circuit is provided wherein (i) the output voltage can be a fraction of the silicon bandgap voltage, (ii) the output voltage can have a zero TC, and (iii) the operating supply voltage can be less than 1.5 V.
In one embodiment of the present invention, the prior art Brokaw bandgap circuit of FIG. 1 is modified so that the operating supply voltage Vcc is lowered together with the output voltage by a constant offset. Referring to FIG. 3, the offset is created using an additional npn bipolar junction transistor (Q2), an opamp (A3) and a plurality of resistors (R5, R6 and R7).
In further embodiments of the present invention, the prior art bandgap reference circuit of FIG. 2 is modified so that the operating supply voltage is lowered together with the output voltage by a constant offset. In one embodiment, referring to FIG. 4, the offset is created using an additional current source I6, NMOS transistor M3, opamp A4, and resistors R8-R10. In another embodiment the offset is created, referring to FIG. 5, by modifying FIG. 4 to omit current source I6, and the resistor R4 shown connected in FIG. 4 is moved to the emitter of transistor Q5.